The company has released its Decisions API model for general use, positioning the service as a quicker and cheaper option compared to conventional large language models.
OpenAI has unveiled its Decisions API, a new model designed for rapid, low‑cost inference, aiming to give developers a cheaper alternative to traditional large language models.
What is the Decisions API?
The Decisions API is built to handle binary or multi‑choice tasks such as classification, ranking, and recommendation. Unlike full‑scale generative models, it focuses on delivering a single, deterministic output, which reduces compute requirements and latency.
Performance and pricing
OpenAI claims the service can process requests up to three times faster than its flagship GPT‑4 model while costing roughly a third of the price per token. Early benchmarks show comparable accuracy on standard decision‑making datasets, though the model is not intended for creative text generation.
- Up to 3× faster inference
- ~33% of GPT‑4 pricing
- Optimized for classification and ranking
Target use cases
Developers building recommendation engines, content moderation pipelines, or real‑time decision support systems can integrate the Decisions API to cut costs without sacrificing reliability.
The API is accessible through the same OpenAI platform used for its other models, with familiar authentication and rate‑limit controls, making adoption straightforward for existing customers.
For more context on OpenAI’s broader strategy, see the OpenAI on Momentum page and explore related stories in the AI category.